This study complied with the principles of the Declaration of Helsinki. The Institutional Review Board (IRB) at at the University of California, San Diego (IRB no., 161794) reviewed and approved the study’s protocol. Informed consent was waived.

This retrospective single-center series included 86 consecutive cases of OLIF performed by a single attending surgeon (senior author M.H.P.). Demographic and perioperative data were recorded, including age, sex, body mass index (BMI) categories, smoking status, previous abdominal surgery, estimated blood loss, operative level, number of levels, spacer dimensions, hospital length of stay, bone graft type, and reoperation (Table 1). Radiographic parameters, including bone mineral density (Hounsfield units), posterior DH, disk angle (DA), FH, SL, lumbar lordosis (LL), and implant subsidence were then analyzed preoperatively and at three postoperative time points, including immediate postoperatively, 3 months, and the most recent or longest follow-up (mean 387 days). The 3-month follow-up data point included radiographs collected closest to 90 days postoperatively, and the most recent follow-up time point was gathered using the latest radiograph obtained. The minimum follow-up for study inclusion was 100 days or at least two available postoperative radiographs. Hounsfield units were measured at L1 on sagittal preoperative computed tomography (CT) scans within the L1 vertebral body boundaries without including the cortical wall. The intervertebral posterior DH was measured between the endplates in the posterior third on the lateral plane. The DA was measured from the inferior endplate of the cranial vertebral body to the superior endplate of the caudal vertebral body. The FH was measured from the inferior pedicle wall of the superior level to the superior pedicle wall of the inferior level. SL was measured from the superior endplate of the cranial level to the inferior endplate of the caudal level, whereas LL was measured from the superior endplate of L1 to the superior endplate of S1 (Fig. 1). Implant subsidence was evaluated based on the Marchi criteria as the presence of vertebral body collapses around the disk space from standing neutral lateral radiographs. In this study, any level with grade ≥I subsidence, defined as >25% loss of postoperative DH, or clear endplate collapse postoperatively was counted as an incidence of subsidence (Fig. 2) [14].

Patients were placed in the left lateral decubitus position, and lateral and paramedial incisions were made under fluoroscopic guidance. The abdominal muscles are dissected longitudinally, and the peritoneal contents are retracted. Once in the retroperitoneal space, the psoas muscle is determined, and the connections with the lateral disk space are removed through careful dissection. Dilators were placed with posterior dissection of the psoas muscle and medial retraction of the retroperitoneal vessels, thereby exposing the lateral circumference of the intervertebral disk and establishing the surgical window. Fluoroscopic imaging was conducted to verify the retractor placement and level. Once verified, a discectomy is performed, followed by spacer placement to enable gradual disk space expansion. Static cages were then prefilled with the bone graft and placed in the disk space, whereas expandable cages were placed in a collapsed configuration, opened in the disk space, and backfilled with the bone graft. The expandable cages utilized in this study were manufactured using titanium alloy and radiolucent polyether-ether-ketone (PEEK) polymer. The static cages were created with a radiolucent PEEK polymer.

MATLAB (Mathworks Inc., Natick, MA, USA) was used for statistical analyses. The independent sample t-test was conducted to assess baseline differences between the cage groups, including continuous demographic variables, reoperation rates, and time to follow-up. The chi-square test was used to evaluate categorical variables. Paired sample t-tests were employed to calculate changes between preoperative and postoperative time points for all interval variables, including posterior DH, LL, SL, FH, and DA, within the cage groups. Multivariable linear regression was conducted to predict the change in all radiographic measures while including demographic factors that popularly affect the radiologic measures collected during this study, particularly age [15], sex (female), BMI [16], bone mineral density as measured with degraded (<135) and non-degraded (≥135) Hounsfield units [17–19], length of follow-up, smoking, number of levels (categorized into 1 level vs. >1 level), spacer dimensions (including width and length), preoperative baseline radiographic measurements and previous abdominal surgery in the model. Subsidence incidence was assessed with a logistic regression model, including all the above predictor variables. All analyses were conducted by level immediately postoperative, 3-month, and recent time points. A p-value of <0.05 indicated statistical significance.

In this study, 86 patients who underwent OLIF at 128 levels were divided into expandable (39 patients, 63 levels) and static (47 patients, 66 levels) cage groups. The enrolled patients included 49 women and 37 men with a mean age of 63.8±11.5 years (range, 31–88 years). Most patients underwent single-level fusion (68.6%, 59/86), whereas an additional 19.8% (17/86) underwent two-level fusion, and 11.6% (10/86) had >3 levels of interbody fusion. L4–L5 level fusion was the most predominant for both the static and expandable groups (51.5% and 47.6%, respectively). No significant differences in any demographic variables or recorded baseline radiographic measurements except for the mean length of follow-up at the most recent time point were found between the static and expandable groups (Tables 1, 2).

The incidence of radiographic subsidence was significantly higher in the static group than in the expandable group (Fig. 3). Cage type predicted the incidence of subsidence. Expandable cages were associated with 4.00 and 2.43 fewer subsidence incidences compared with static cages at the postoperative and most recent time points (p<0.05).

Patients were assessed for reoperation. Each patient who underwent subsequent revision surgery at the same level was included. The static and expandable groups involved five and two cases of reoperation, respectively (p>0.05). Refractory pain after the procedure was the most cited cause for reoperation, which was present in 4/5 patients in the static group and none in the expandable group. Other reasons for reoperation included wound revision in two patients, one in both the static and expandable groups, as well as persistent radiculopathy in one patient in the expandable group. Subsidence was present on radiographs before reoperation in 3/5 patients in the static group and none in the expandable group.

We then analyzed the additional radiographic measures within the cage groups and observed significant increases in posterior DH, FH, and DA within the static group and posterior DH, FH, SL, LL, and DA within the expandable group (Supplement 1). To assess between-group differences, we used a linear regression model to predict the change in each radiographic measure compared with the baseline measurement for each time point. In both the posterior disk and FH models, cage type was a significant predictor of the change in height. Expandable cages were associated with improved posterior DH expansion at all three time points when compared with the static cage type (1.24 mm, 0.88 mm, and 1.85 mm; p<0.01). Expandable cages in the FH model were associated with larger FH increases at 3 months and the most recent follow-up (1.12 mm, 0.40 mm, and 1.28 mm; p=0.096, 0.016, and 0.030). The expandable cage type was associated with significant improvement (3.46°, 3.12°, and 3.36°; p<0.01, 0.05, and 0.08) at the immediate postoperative and 3-month time points compared with the static group when predicting the change in SL (Figs. 4, 5). No significant differences in terms of changes in DA and LL were found between the cage groups (Fig. 6).